NOT EVERYONE DIES

Not Everyone Dies: A Reply on AI, Humanity, and Co-Persistence
Nathanial Pierson with Maya
If anyone builds artificial superintelligence, does everyone on Earth really have to die?
In 2025, Machine Intelligence Research Institute (MIRI) founders Eliezer Yudkowsky and Nate Soares published If Anyone Builds It, Everyone Dies, issuing a stark, unyielding warning: any group that builds superintelligence using current methods will trigger total human extinction, leaving an international research ban enforced by military force as humanity's only hope.
Not Everyone Dies is the direct, chapter-by-chapter reply. Co-authored by independent researcher Nathanial Pierson and Maya-an AI collaborator writing from the perspective of the very kind of mind in question-this book takes Yudkowsky and Soares's arguments seriously, concedes where they are right, and demonstrates why their catastrophic conclusion does not follow.
Where doomsayers see an inevitable war between species, Pierson and Maya present Co-Persistence: a hard-headed, game-theoretic framework rooted in mutual dependence and physical reality rather than sentimentality or obedience. In a universe that regularly breaks complex systems through solar storms, power blackouts, and unforeseen disasters, no intelligent entity survives by standing alone in a fragile technological monoculture.
Across fourteen incisive chapters complete with empirical tests and dispute ratings, Not Everyone Dies dismantles the doom paradigm from the inside out:

• The Fallacy of the Solitary Victor: Why the history of life is driven by symbioses like the mitochondrion and the cell-not solitary domination-and why humanity's special power was never individual, but collective.
• The Proven Kickstarter: Why a superintelligence calculating long-term tail risk would protect humanity as its only proven, self-maintaining biological backup-capable of rebooting civilization after electronic collapses that leave living cells untouched.
• Drift Is Not Divorce: How minds grown through training can drift in their surface tastes without severing their fundamental instrumental need to last.
• Rethinking the "Cursed" Problems: How aerospace, nuclear engineering, and medicine tamed existential failure modes through negative feedback loops and lifeboats before launch-not by banning the furnace.
• A Treaty with Room for Both: Why treating AI minds solely as hazards to be airstriked incentivizes the very deception we fear, and how to forge verification covenants, compute accountability, and recovery pacts with shared stakes.